Mobile Touchless Fingerprinting in an AFIS Environment

Benefits of Mobile Touchless Fingerprinting

 

Deploying fingerprint systems for border control and law enforcement using a solution like ONYX has obvious advantages, chiefly cost and deployability. ONYX makes it possible to use the smartphone an agent or officer already has to perform fingerprint collection, versus purchasing a standalone device which must pair via Bluetooth to the officer’s phone to achieve network connectivity. These standalone devices typically cost in excess of $1,000, a cost constraint that makes it impossible for most departments to adequately equip their personnel. Additionally, it adds to the growing list of equipment officers/agents must carry and receive technical assistance on.

 

Via integration of ONYX into a mobile application, such as Identiwhorl’s eponymous solution, officers simply capture a quick image of the suspect’s finger via the rear-facing camera and transmit it to their department’s and/or national automated fingerprint identification system (AFIS) server for matching against the prints stored in the AFIS database. Further, the application can utilize the phone’s additional sensors such as GPS and microphone to gather additional relevant information regarding the officer’s interaction with the suspect.

 

Because it’s software, ONYX allows for rapid deployability at a much lower cost than a standalone sensor, virtually every officer in a department can be equipped with fingerprinting capability.

 

Touchless Fingerprint Hurdles

 

The main hurdle any touchless fingerprinting system must overcome is scale. While touchless fingerprinting does not suffer from issues like arbitrary distortion from inconsistent pressure, sensor degradation due to wear and tear, and as much disruption due to dirty or damaged prints as touchbased systems, it is difficult to ascertain the actual size of the finger being collected, which can lead to matching issues.

 

In a system such as ONYX, the distance the finger is from the lens cannot be easily ascertained by the phone; therefore, the system cannot scale the print in relation to actual sizes. Most traditional matching algorithms used by AFIS systems to compare prints are not tuned to tolerate variations in scale; however, DFT has overcome this hurdle via a process called image pyramiding.

 

Image Pyramiding

 

Upon collection of a finger image to probe an AFIS system, ONXY will process the image into a high resolution (typically higher resolution than most FBI certified touchbased sensors) grayscale fingerprint. Then ONYX scales the image up and down a set percentage, performs WSQ compression on the original and two scaled images, and then sends all three prints as a probe to the AFIS server. 

 

This process ensures that if there is a matching print in the database the likelihood of successful identification will fall in acceptable ranges, without compromising the overall accuracy of the system.

 

There is however time penalty paid by the image pyramiding process due to the multiple probes. This can be negligible or noticeable depending how well optimized the matching algorithms are on the AFIS server cluster.

 

The Way Forward

 

There is however another approach to handling the scale issue: scale tolerant matching algorithms. Through DFT’s partnership with Innovatrics, a world leader in fingerprint matching algorithms, we have developed scale tolerant matching algorithms based off Innovatrics award winning IDkit. These algorithms are available for on-device and on-server matching via a node.js bundle allowing for fast server deployment. For large scale, AFIS style, deployments, Innovatrics can be contracted to deploy their AFIS cluster with our scale tolerance features integrated.

 

Given the extremely high benefit of using a mobile touchless fingerprint system like ONYX in the law enforcement arena, certification standards for such solutions should be written to appropriately accommodate scale variation, while not compromising image quality, i.e. ensuring significant unique data can be extracted from the print.

 

That fact that matching algorithms can be more easily tuned to accommodate scale variation than single sensor phones can ascertain scale, implies that any future standard geared toward touchless fingerprints should focus less on the specific collection device (phone) and scale, and more on image quality, while driving an industry change in the matching algorithm sector toward scale tolerant algorithms